All articles
Topic guide
Production AI architecture
Getting an AI feature to work in a demo is the easy part. These articles cover what surrounds the model once real users depend on it: how requests are routed and validated, how failures are contained, how agents and tools are wired together, and what to check before launch.
Start with the request path if you are designing a new system, with reliability if something already runs in production, and with the tools below if you need numbers for a decision.
The request path and gateways
How a request travels from the user to the model and back, and where to put the controls.
- Designing the AI Request Pipeline: 8 Layers Between User Input and Your LLMA production architecture blueprint for every layer between a user's message and a safe, reliable LLM response — with Spring Boot code you can actually use.9 min read
- AI Gateways: Managing LLM Traffic in the EnterpriseAs LLM usage spreads across an org, you need a control point. What an AI gateway is, the problems it solves, and the capabilities to look for.3 min read
- Building an AI Gateway with Apigee (LLM Proxy)Put Apigee in front of your LLM providers to centralize API keys, cost control, quotas, logging, and guardrails — a blueprint for an enterprise AI gateway.4 min read
- Spring AI — Production Architecture & Best PracticesTake Spring AI to production: a reference architecture plus observability, caching, resilience, evals, security, and the anti-patterns to avoid.2 min read
Agents, tools and memory
Designing agents that call tools, share context and remember, without losing control of them.
- Building Enterprise AI AgentsArchitecture patterns and best practices for building production-grade AI agents that scale — from orchestration to observability.6 min read
- Agentic AI Explained: From Prompt to Autonomous SystemsWhat agentic AI actually is, how the agent decision loop works, single vs multi-agent systems, the enterprise stack, real use cases, and the risks nobody skips.8 min read
- I Rebuilt Our AI Agents on Kafka Instead of REST — Here's the Full Java ArchitectureREST between AI agents works until agent #4 shows up. The full Java/Spring Kafka topic design, the ReAct loop as a state machine, and a free audit trail.10 min read
- MCP vs Function Calling: What's the Difference (and When to Use Each)Model Context Protocol and function calling both give LLMs access to tools, but solve different problems — a clear comparison and guidance on when to use each.4 min read
- Building an MCP Server with Spring AIMost MCP content is Python/TypeScript. Here's how to build an MCP server in Java with Spring AI — expose your services as tools any MCP client can use.2 min read
- Agent Memory Patterns: How to Give Your LLM Application a Brain That PersistsYour LLM forgets everything once the request ends. How production systems build short-term, long-term, and semantic memory — with Spring AI and pgvector.6 min read
- Context Engineering: The Real Skill Behind Reliable LLM AppsWhat you put in the context window matters more than prompt wording. A practical guide to context engineering — the budget, techniques, and failure modes.3 min read
Reliability, rollout and launch
Keeping AI features working when the model, the network or the data misbehaves, and shipping them safely.
- Building Fault-Tolerant AI SystemsResilience patterns for production AI — circuit breakers, fallback chains, and graceful degradation so systems survive provider outages and rate limits.3 min read
- Event-Driven AI: How Kafka Streams Changed the Way We Run LLMs in ProductionFrom blocking HTTP calls to async event streams — the architecture shift that halved latency, eliminated 3AM pages, and lets us process 2M AI requests a day.11 min read
- Feature Flags for AI: The Deployment Pattern That Saved Us From 3 Production DisastersWhy LLM features need flags more than anything you've shipped — the four flag patterns that let us swap models, A/B test prompts, and kill broken features fast.8 min read
- Why Most Enterprise AI Projects Die in Production (And the 5 Patterns That Actually Survive)I've deployed AI to 10,000+ enterprise users across healthcare, logistics, and finance. Failures weren't about the model — they happened earlier.9 min read
- Zero to Production: Building Your First Enterprise LLM ApplicationA four-phase guide to taking an LLM prototype to a production enterprise app — RAG, caching, observability, cost control, and multi-model routing.6 min read
- 5 Security Checks Every App Builder Must Run Before LaunchA practical checklist of 5 AI-assisted security audits — secrets, data flow, production hardening, deep vulnerabilities, and the attacker's perspective.6 min read
Free tools for this topic
Free calculators and checkers that go with these articles.
- Enterprise AI Readiness AssessmentScore your organization's AI readiness across 8 dimensions in 2 minutes — and get the top focus areas to close the gaps.
- Which Claude Model Should I Use?Answer three quick questions and get a model-tier recommendation with the reasoning.
- LLM Cost EstimatorEstimate the cost per request and the total for your request volume, from token counts and per-million-token pricing — input and output.
Preparing for a Claude certification instead? See the Claude certification hub.